net-next-2.6 reverts.... PLEASE READ

net-next-2.6 reverts.... PLEASE READ

[David Miller]

I had to pop off the top 3 commits in net-next-2.6 in order to fix
them up and get rid of some very unintentional changes that slipped
in.

Some defconfig files got modified in the NAPI config option removal
commits that went in yesterday, and that's not supposed to happen.

If you've already pulled, you can fix your net-next-2.6 tree like
this:

1) Put "38b221957b155ef410bdc28856a66386303fbd5a" into the
   .git/refs/heads/master file.

2) Repull.

I hope this doesn't inconvenience too many people.

I didn't want to even think about rebuilding the entire tree
just to fix this. :-/

Re: net-next-2.6 reverts.... PLEASE READ

[Francois Romieu]

David Miller <davem@davemloft.net> :
[...]
> Some defconfig files got modified in the NAPI config option removal
> commits that went in yesterday, and that's not supposed to happen.

Is it a matter of keeping the option for a longer time or should they
not go away under any circumstance ?

-- 
Ueimor

Re: net-next-2.6 reverts.... PLEASE READ

[David Miller]

From: Francois Romieu <romieu@fr.zoreil.com>
Date: Tue, 8 Jul 2008 09:01:06 +0200

> David Miller <davem@davemloft.net> :
> [...]
> > Some defconfig files got modified in the NAPI config option removal
> > commits that went in yesterday, and that's not supposed to happen.
> 
> Is it a matter of keeping the option for a longer time or should they
> not go away under any circumstance ?

The issue is that defconfig updates should stay out of the
subsystem trees.

Otherwise we get conflicts with the arch maintainer trees,
and also "make oldconfig" works perfectly fine without
the defconfig change.

Re: net-next-2.6 reverts.... PLEASE READ

[John W. Linville]

On Mon, Jul 07, 2008 at 10:43:36PM -0700, David Miller wrote:

> If you've already pulled, you can fix your net-next-2.6 tree like
> this:
> 
> 1) Put "38b221957b155ef410bdc28856a66386303fbd5a" into the
>    .git/refs/heads/master file.

This part was especially helpful...thanks! :-)

John
-- 
John W. Linville
linville@tuxdriver.com

[PATCH 1/6] cxgb: delete non NAPI code from the driver.

[Francois Romieu]

Compile-tested only.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 drivers/net/Kconfig         |    8 -----
 drivers/net/chelsio/cxgb2.c |    2 -
 drivers/net/chelsio/sge.c   |   70 +++---------------------------------------
 3 files changed, 5 insertions(+), 75 deletions(-)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 4675c1b..50ca1cf 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2378,14 +2378,6 @@ config CHELSIO_T1_1G
           Enables support for Chelsio's gigabit Ethernet PCI cards.  If you
           are using only 10G cards say 'N' here.
 
-config CHELSIO_T1_NAPI
-	bool "Use Rx Polling (NAPI)"
-	depends on CHELSIO_T1
-	default y
-	help
-	  NAPI is a driver API designed to reduce CPU and interrupt load
-	  when the driver is receiving lots of packets from the card.
-
 config CHELSIO_T3
 	tristate "Chelsio Communications T3 10Gb Ethernet support"
 	depends on PCI && INET
diff --git a/drivers/net/chelsio/cxgb2.c b/drivers/net/chelsio/cxgb2.c
index a509337..638c9a2 100644
--- a/drivers/net/chelsio/cxgb2.c
+++ b/drivers/net/chelsio/cxgb2.c
@@ -1153,9 +1153,7 @@ static int __devinit init_one(struct pci_dev *pdev,
 #ifdef CONFIG_NET_POLL_CONTROLLER
 		netdev->poll_controller = t1_netpoll;
 #endif
-#ifdef CONFIG_CHELSIO_T1_NAPI
 		netif_napi_add(netdev, &adapter->napi, t1_poll, 64);
-#endif
 
 		SET_ETHTOOL_OPS(netdev, &t1_ethtool_ops);
 	}
diff --git a/drivers/net/chelsio/sge.c b/drivers/net/chelsio/sge.c
index 8a7efd3..d6c7d2a 100644
--- a/drivers/net/chelsio/sge.c
+++ b/drivers/net/chelsio/sge.c
@@ -1396,20 +1396,10 @@ static void sge_rx(struct sge *sge, struct freelQ *fl, unsigned int len)
 
 	if (unlikely(adapter->vlan_grp && p->vlan_valid)) {
 		st->vlan_xtract++;
-#ifdef CONFIG_CHELSIO_T1_NAPI
-			vlan_hwaccel_receive_skb(skb, adapter->vlan_grp,
-						 ntohs(p->vlan));
-#else
-			vlan_hwaccel_rx(skb, adapter->vlan_grp,
-					ntohs(p->vlan));
-#endif
-	} else {
-#ifdef CONFIG_CHELSIO_T1_NAPI
+		vlan_hwaccel_receive_skb(skb, adapter->vlan_grp,
+					 ntohs(p->vlan));
+	} else
 		netif_receive_skb(skb);
-#else
-		netif_rx(skb);
-#endif
-	}
 }
 
 /*
@@ -1568,7 +1558,6 @@ static inline int responses_pending(const struct adapter *adapter)
 	return (e->GenerationBit == Q->genbit);
 }
 
-#ifdef CONFIG_CHELSIO_T1_NAPI
 /*
  * A simpler version of process_responses() that handles only pure (i.e.,
  * non data-carrying) responses.  Such respones are too light-weight to justify
@@ -1636,9 +1625,6 @@ int t1_poll(struct napi_struct *napi, int budget)
 	return work_done;
 }
 
-/*
- * NAPI version of the main interrupt handler.
- */
 irqreturn_t t1_interrupt(int irq, void *data)
 {
 	struct adapter *adapter = data;
@@ -1656,7 +1642,8 @@ irqreturn_t t1_interrupt(int irq, void *data)
 			else {
 				/* no data, no NAPI needed */
 				writel(sge->respQ.cidx, adapter->regs + A_SG_SLEEPING);
-				napi_enable(&adapter->napi);	/* undo schedule_prep */
+				/* undo schedule_prep */
+				napi_enable(&adapter->napi);
 			}
 		}
 		return IRQ_HANDLED;
@@ -1672,53 +1659,6 @@ irqreturn_t t1_interrupt(int irq, void *data)
 	return IRQ_RETVAL(handled != 0);
 }
 
-#else
-/*
- * Main interrupt handler, optimized assuming that we took a 'DATA'
- * interrupt.
- *
- * 1. Clear the interrupt
- * 2. Loop while we find valid descriptors and process them; accumulate
- *      information that can be processed after the loop
- * 3. Tell the SGE at which index we stopped processing descriptors
- * 4. Bookkeeping; free TX buffers, ring doorbell if there are any
- *      outstanding TX buffers waiting, replenish RX buffers, potentially
- *      reenable upper layers if they were turned off due to lack of TX
- *      resources which are available again.
- * 5. If we took an interrupt, but no valid respQ descriptors was found we
- *      let the slow_intr_handler run and do error handling.
- */
-irqreturn_t t1_interrupt(int irq, void *cookie)
-{
-	int work_done;
-	struct adapter *adapter = cookie;
-	struct respQ *Q = &adapter->sge->respQ;
-
-	spin_lock(&adapter->async_lock);
-
-	writel(F_PL_INTR_SGE_DATA, adapter->regs + A_PL_CAUSE);
-
-	if (likely(responses_pending(adapter)))
-		work_done = process_responses(adapter, -1);
-	else
-		work_done = t1_slow_intr_handler(adapter);
-
-	/*
-	 * The unconditional clearing of the PL_CAUSE above may have raced
-	 * with DMA completion and the corresponding generation of a response
-	 * to cause us to miss the resulting data interrupt.  The next write
-	 * is also unconditional to recover the missed interrupt and render
-	 * this race harmless.
-	 */
-	writel(Q->cidx, adapter->regs + A_SG_SLEEPING);
-
-	if (!work_done)
-		adapter->sge->stats.unhandled_irqs++;
-	spin_unlock(&adapter->async_lock);
-	return IRQ_RETVAL(work_done != 0);
-}
-#endif
-
 /*
  * Enqueues the sk_buff onto the cmdQ[qid] and has hardware fetch it.
  *
-- 
1.5.3.3

[PATCH 2/6] via-rhine: delete non NAPI code from the driver.

[Francois Romieu]

Compile-tested only.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 drivers/net/Kconfig     |   11 -----------
 drivers/net/via-rhine.c |   27 ++-------------------------
 2 files changed, 2 insertions(+), 36 deletions(-)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 50ca1cf..fb618b6 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -1694,17 +1694,6 @@ config VIA_RHINE_MMIO
 
 	  If unsure, say Y.
 
-config VIA_RHINE_NAPI
-	bool "Use Rx Polling (NAPI)"
-	depends on VIA_RHINE
-	help
-	  NAPI is a new driver API designed to reduce CPU and interrupt load
-	  when the driver is receiving lots of packets from the card.
-
-	  If your estimated Rx load is 10kpps or more, or if the card will be
-	  deployed on potentially unfriendly networks (e.g. in a firewall),
-	  then say Y here.
-
 config LAN_SAA9730
 	bool "Philips SAA9730 Ethernet support"
 	depends on NET_PCI && PCI && MIPS_ATLAS
diff --git a/drivers/net/via-rhine.c b/drivers/net/via-rhine.c
index 8c9d6ae..96dff04 100644
--- a/drivers/net/via-rhine.c
+++ b/drivers/net/via-rhine.c
@@ -73,12 +73,7 @@ static const int multicast_filter_limit = 32;
    There are no ill effects from too-large receive rings. */
 #define TX_RING_SIZE	16
 #define TX_QUEUE_LEN	10	/* Limit ring entries actually used. */
-#ifdef CONFIG_VIA_RHINE_NAPI
 #define RX_RING_SIZE	64
-#else
-#define RX_RING_SIZE	16
-#endif
-
 
 /* Operational parameters that usually are not changed. */
 
@@ -583,7 +578,6 @@ static void rhine_poll(struct net_device *dev)
 }
 #endif
 
-#ifdef CONFIG_VIA_RHINE_NAPI
 static int rhine_napipoll(struct napi_struct *napi, int budget)
 {
 	struct rhine_private *rp = container_of(napi, struct rhine_private, napi);
@@ -604,7 +598,6 @@ static int rhine_napipoll(struct napi_struct *napi, int budget)
 	}
 	return work_done;
 }
-#endif
 
 static void __devinit rhine_hw_init(struct net_device *dev, long pioaddr)
 {
@@ -784,9 +777,8 @@ static int __devinit rhine_init_one(struct pci_dev *pdev,
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	dev->poll_controller = rhine_poll;
 #endif
-#ifdef CONFIG_VIA_RHINE_NAPI
 	netif_napi_add(dev, &rp->napi, rhine_napipoll, 64);
-#endif
+
 	if (rp->quirks & rqRhineI)
 		dev->features |= NETIF_F_SG|NETIF_F_HW_CSUM;
 
@@ -1056,9 +1048,7 @@ static void init_registers(struct net_device *dev)
 
 	rhine_set_rx_mode(dev);
 
-#ifdef CONFIG_VIA_RHINE_NAPI
 	napi_enable(&rp->napi);
-#endif
 
 	/* Enable interrupts by setting the interrupt mask. */
 	iowrite16(IntrRxDone | IntrRxErr | IntrRxEmpty| IntrRxOverflow |
@@ -1193,9 +1183,7 @@ static void rhine_tx_timeout(struct net_device *dev)
 	/* protect against concurrent rx interrupts */
 	disable_irq(rp->pdev->irq);
 
-#ifdef CONFIG_VIA_RHINE_NAPI
 	napi_disable(&rp->napi);
-#endif
 
 	spin_lock(&rp->lock);
 
@@ -1319,16 +1307,12 @@ static irqreturn_t rhine_interrupt(int irq, void *dev_instance)
 
 		if (intr_status & (IntrRxDone | IntrRxErr | IntrRxDropped |
 				   IntrRxWakeUp | IntrRxEmpty | IntrRxNoBuf)) {
-#ifdef CONFIG_VIA_RHINE_NAPI
 			iowrite16(IntrTxAborted |
 				  IntrTxDone | IntrTxError | IntrTxUnderrun |
 				  IntrPCIErr | IntrStatsMax | IntrLinkChange,
 				  ioaddr + IntrEnable);
 
 			netif_rx_schedule(dev, &rp->napi);
-#else
-			rhine_rx(dev, RX_RING_SIZE);
-#endif
 		}
 
 		if (intr_status & (IntrTxErrSummary | IntrTxDone)) {
@@ -1520,11 +1504,7 @@ static int rhine_rx(struct net_device *dev, int limit)
 						 PCI_DMA_FROMDEVICE);
 			}
 			skb->protocol = eth_type_trans(skb, dev);
-#ifdef CONFIG_VIA_RHINE_NAPI
 			netif_receive_skb(skb);
-#else
-			netif_rx(skb);
-#endif
 			dev->last_rx = jiffies;
 			rp->stats.rx_bytes += pkt_len;
 			rp->stats.rx_packets++;
@@ -1836,9 +1816,7 @@ static int rhine_close(struct net_device *dev)
 	spin_lock_irq(&rp->lock);
 
 	netif_stop_queue(dev);
-#ifdef CONFIG_VIA_RHINE_NAPI
 	napi_disable(&rp->napi);
-#endif
 
 	if (debug > 1)
 		printk(KERN_DEBUG "%s: Shutting down ethercard, "
@@ -1937,9 +1915,8 @@ static int rhine_suspend(struct pci_dev *pdev, pm_message_t state)
 	if (!netif_running(dev))
 		return 0;
 
-#ifdef CONFIG_VIA_RHINE_NAPI
 	napi_disable(&rp->napi);
-#endif
+
 	netif_device_detach(dev);
 	pci_save_state(pdev);
 
-- 
1.5.3.3

[PATCH 3/6] e1000: delete non NAPI code from the driver.

[Francois Romieu]

Compile-tested only.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 Documentation/networking/e1000.txt |   14 +---
 drivers/net/Kconfig                |   14 ----
 drivers/net/e1000/e1000.h          |   11 +--
 drivers/net/e1000/e1000_main.c     |  150 +++--------------------------------
 4 files changed, 17 insertions(+), 172 deletions(-)

diff --git a/Documentation/networking/e1000.txt b/Documentation/networking/e1000.txt
index 61b171c..2df7186 100644
--- a/Documentation/networking/e1000.txt
+++ b/Documentation/networking/e1000.txt
@@ -513,21 +513,11 @@ Additional Configurations
   Intel(R) PRO/1000 PT Dual Port Server Connection
   Intel(R) PRO/1000 PT Dual Port Server Adapter
   Intel(R) PRO/1000 PF Dual Port Server Adapter
-  Intel(R) PRO/1000 PT Quad Port Server Adapter 
+  Intel(R) PRO/1000 PT Quad Port Server Adapter
 
   NAPI
   ----
-  NAPI (Rx polling mode) is supported in the e1000 driver.  NAPI is enabled
-  or disabled based on the configuration of the kernel.  To override
-  the default, use the following compile-time flags.
-
-  To enable NAPI, compile the driver module, passing in a configuration option:
-
-       make CFLAGS_EXTRA=-DE1000_NAPI install
-
-  To disable NAPI, compile the driver module, passing in a configuration option:
-
-       make CFLAGS_EXTRA=-DE1000_NO_NAPI install
+  NAPI (Rx polling mode) is enabled in the e1000 driver.
 
   See www.cyberus.ca/~hadi/usenix-paper.tgz for more information on NAPI.
 
diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index fb618b6..33b8899 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -1945,20 +1945,6 @@ config E1000
 	  To compile this driver as a module, choose M here. The module
 	  will be called e1000.
 
-config E1000_NAPI
-	bool "Use Rx Polling (NAPI)"
-	depends on E1000
-	help
-	  NAPI is a new driver API designed to reduce CPU and interrupt load
-	  when the driver is receiving lots of packets from the card. It is
-	  still somewhat experimental and thus not yet enabled by default.
-
-	  If your estimated Rx load is 10kpps or more, or if the card will be
-	  deployed on potentially unfriendly networks (e.g. in a firewall),
-	  then say Y here.
-
-	  If in doubt, say N.
-
 config E1000_DISABLE_PACKET_SPLIT
 	bool "Disable Packet Split for PCI express adapters"
 	depends on E1000
diff --git a/drivers/net/e1000/e1000.h b/drivers/net/e1000/e1000.h
index 31feae1..dfc5b39 100644
--- a/drivers/net/e1000/e1000.h
+++ b/drivers/net/e1000/e1000.h
@@ -246,9 +246,7 @@ struct e1000_adapter {
 	u16 link_speed;
 	u16 link_duplex;
 	spinlock_t stats_lock;
-#ifdef CONFIG_E1000_NAPI
 	spinlock_t tx_queue_lock;
-#endif
 	unsigned int total_tx_bytes;
 	unsigned int total_tx_packets;
 	unsigned int total_rx_bytes;
@@ -286,22 +284,17 @@ struct e1000_adapter {
 	bool detect_tx_hung;
 
 	/* RX */
-#ifdef CONFIG_E1000_NAPI
 	bool (*clean_rx) (struct e1000_adapter *adapter,
 			  struct e1000_rx_ring *rx_ring,
 			  int *work_done, int work_to_do);
-#else
-	bool (*clean_rx) (struct e1000_adapter *adapter,
-			  struct e1000_rx_ring *rx_ring);
-#endif
 	void (*alloc_rx_buf) (struct e1000_adapter *adapter,
 			      struct e1000_rx_ring *rx_ring,
 				int cleaned_count);
 	struct e1000_rx_ring *rx_ring;      /* One per active queue */
-#ifdef CONFIG_E1000_NAPI
+
 	struct napi_struct napi;
 	struct net_device *polling_netdev;  /* One per active queue */
-#endif
+
 	int num_tx_queues;
 	int num_rx_queues;
 
diff --git a/drivers/net/e1000/e1000_main.c b/drivers/net/e1000/e1000_main.c
index 7aa458f..ead4071 100644
--- a/drivers/net/e1000/e1000_main.c
+++ b/drivers/net/e1000/e1000_main.c
@@ -31,12 +31,7 @@
 
 char e1000_driver_name[] = "e1000";
 static char e1000_driver_string[] = "Intel(R) PRO/1000 Network Driver";
-#ifndef CONFIG_E1000_NAPI
-#define DRIVERNAPI
-#else
-#define DRIVERNAPI "-NAPI"
-#endif
-#define DRV_VERSION "7.3.20-k2"DRIVERNAPI
+#define DRV_VERSION "7.3.20-k3-NAPI"
 const char e1000_driver_version[] = DRV_VERSION;
 static const char e1000_copyright[] = "Copyright (c) 1999-2006 Intel Corporation.";
 
@@ -138,7 +133,6 @@ static irqreturn_t e1000_intr(int irq, void *data);
 static irqreturn_t e1000_intr_msi(int irq, void *data);
 static bool e1000_clean_tx_irq(struct e1000_adapter *adapter,
 			       struct e1000_tx_ring *tx_ring);
-#ifdef CONFIG_E1000_NAPI
 static int e1000_clean(struct napi_struct *napi, int budget);
 static bool e1000_clean_rx_irq(struct e1000_adapter *adapter,
 			       struct e1000_rx_ring *rx_ring,
@@ -146,12 +140,6 @@ static bool e1000_clean_rx_irq(struct e1000_adapter *adapter,
 static bool e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
 				  struct e1000_rx_ring *rx_ring,
 				  int *work_done, int work_to_do);
-#else
-static bool e1000_clean_rx_irq(struct e1000_adapter *adapter,
-			       struct e1000_rx_ring *rx_ring);
-static bool e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
-				  struct e1000_rx_ring *rx_ring);
-#endif
 static void e1000_alloc_rx_buffers(struct e1000_adapter *adapter,
                                    struct e1000_rx_ring *rx_ring,
 				   int cleaned_count);
@@ -511,9 +499,8 @@ int e1000_up(struct e1000_adapter *adapter)
 
 	clear_bit(__E1000_DOWN, &adapter->flags);
 
-#ifdef CONFIG_E1000_NAPI
 	napi_enable(&adapter->napi);
-#endif
+
 	e1000_irq_enable(adapter);
 
 	/* fire a link change interrupt to start the watchdog */
@@ -600,9 +587,8 @@ e1000_down(struct e1000_adapter *adapter)
 	 * reschedule our watchdog timer */
 	set_bit(__E1000_DOWN, &adapter->flags);
 
-#ifdef CONFIG_E1000_NAPI
 	napi_disable(&adapter->napi);
-#endif
+
 	e1000_irq_disable(adapter);
 
 	del_timer_sync(&adapter->tx_fifo_stall_timer);
@@ -957,9 +943,7 @@ e1000_probe(struct pci_dev *pdev,
 	e1000_set_ethtool_ops(netdev);
 	netdev->tx_timeout = &e1000_tx_timeout;
 	netdev->watchdog_timeo = 5 * HZ;
-#ifdef CONFIG_E1000_NAPI
 	netif_napi_add(netdev, &adapter->napi, e1000_clean, 64);
-#endif
 	netdev->vlan_rx_register = e1000_vlan_rx_register;
 	netdev->vlan_rx_add_vid = e1000_vlan_rx_add_vid;
 	netdev->vlan_rx_kill_vid = e1000_vlan_rx_kill_vid;
@@ -1199,16 +1183,12 @@ err_eeprom:
 	if (adapter->hw.flash_address)
 		iounmap(adapter->hw.flash_address);
 err_flashmap:
-#ifdef CONFIG_E1000_NAPI
 	for (i = 0; i < adapter->num_rx_queues; i++)
 		dev_put(&adapter->polling_netdev[i]);
-#endif
 
 	kfree(adapter->tx_ring);
 	kfree(adapter->rx_ring);
-#ifdef CONFIG_E1000_NAPI
 	kfree(adapter->polling_netdev);
-#endif
 err_sw_init:
 	iounmap(adapter->hw.hw_addr);
 err_ioremap:
@@ -1236,9 +1216,7 @@ e1000_remove(struct pci_dev *pdev)
 {
 	struct net_device *netdev = pci_get_drvdata(pdev);
 	struct e1000_adapter *adapter = netdev_priv(netdev);
-#ifdef CONFIG_E1000_NAPI
 	int i;
-#endif
 
 	cancel_work_sync(&adapter->reset_task);
 
@@ -1248,10 +1226,8 @@ e1000_remove(struct pci_dev *pdev)
 	 * would have already happened in close and is redundant. */
 	e1000_release_hw_control(adapter);
 
-#ifdef CONFIG_E1000_NAPI
 	for (i = 0; i < adapter->num_rx_queues; i++)
 		dev_put(&adapter->polling_netdev[i]);
-#endif
 
 	unregister_netdev(netdev);
 
@@ -1260,9 +1236,7 @@ e1000_remove(struct pci_dev *pdev)
 
 	kfree(adapter->tx_ring);
 	kfree(adapter->rx_ring);
-#ifdef CONFIG_E1000_NAPI
 	kfree(adapter->polling_netdev);
-#endif
 
 	iounmap(adapter->hw.hw_addr);
 	if (adapter->hw.flash_address)
@@ -1289,9 +1263,7 @@ e1000_sw_init(struct e1000_adapter *adapter)
 	struct e1000_hw *hw = &adapter->hw;
 	struct net_device *netdev = adapter->netdev;
 	struct pci_dev *pdev = adapter->pdev;
-#ifdef CONFIG_E1000_NAPI
 	int i;
-#endif
 
 	/* PCI config space info */
 
@@ -1349,14 +1321,12 @@ e1000_sw_init(struct e1000_adapter *adapter)
 		return -ENOMEM;
 	}
 
-#ifdef CONFIG_E1000_NAPI
 	for (i = 0; i < adapter->num_rx_queues; i++) {
 		adapter->polling_netdev[i].priv = adapter;
 		dev_hold(&adapter->polling_netdev[i]);
 		set_bit(__LINK_STATE_START, &adapter->polling_netdev[i].state);
 	}
 	spin_lock_init(&adapter->tx_queue_lock);
-#endif
 
 	/* Explicitly disable IRQ since the NIC can be in any state. */
 	e1000_irq_disable(adapter);
@@ -1392,7 +1362,6 @@ e1000_alloc_queues(struct e1000_adapter *adapter)
 		return -ENOMEM;
 	}
 
-#ifdef CONFIG_E1000_NAPI
 	adapter->polling_netdev = kcalloc(adapter->num_rx_queues,
 	                                  sizeof(struct net_device),
 	                                  GFP_KERNEL);
@@ -1401,7 +1370,6 @@ e1000_alloc_queues(struct e1000_adapter *adapter)
 		kfree(adapter->rx_ring);
 		return -ENOMEM;
 	}
-#endif
 
 	return E1000_SUCCESS;
 }
@@ -1466,9 +1434,7 @@ e1000_open(struct net_device *netdev)
 	/* From here on the code is the same as e1000_up() */
 	clear_bit(__E1000_DOWN, &adapter->flags);
 
-#ifdef CONFIG_E1000_NAPI
 	napi_enable(&adapter->napi);
-#endif
 
 	e1000_irq_enable(adapter);
 
@@ -2067,11 +2033,9 @@ e1000_configure_rx(struct e1000_adapter *adapter)
 		ctrl_ext = E1000_READ_REG(hw, CTRL_EXT);
 		/* Reset delay timers after every interrupt */
 		ctrl_ext |= E1000_CTRL_EXT_INT_TIMER_CLR;
-#ifdef CONFIG_E1000_NAPI
 		/* Auto-Mask interrupts upon ICR access */
 		ctrl_ext |= E1000_CTRL_EXT_IAME;
 		E1000_WRITE_REG(hw, IAM, 0xffffffff);
-#endif
 		E1000_WRITE_REG(hw, CTRL_EXT, ctrl_ext);
 		E1000_WRITE_FLUSH(hw);
 	}
@@ -3793,9 +3757,6 @@ e1000_intr_msi(int irq, void *data)
 	struct net_device *netdev = data;
 	struct e1000_adapter *adapter = netdev_priv(netdev);
 	struct e1000_hw *hw = &adapter->hw;
-#ifndef CONFIG_E1000_NAPI
-	int i;
-#endif
 	u32 icr = E1000_READ_REG(hw, ICR);
 
 	/* in NAPI mode read ICR disables interrupts using IAM */
@@ -3816,7 +3777,6 @@ e1000_intr_msi(int irq, void *data)
 			mod_timer(&adapter->watchdog_timer, jiffies + 1);
 	}
 
-#ifdef CONFIG_E1000_NAPI
 	if (likely(netif_rx_schedule_prep(netdev, &adapter->napi))) {
 		adapter->total_tx_bytes = 0;
 		adapter->total_tx_packets = 0;
@@ -3825,20 +3785,6 @@ e1000_intr_msi(int irq, void *data)
 		__netif_rx_schedule(netdev, &adapter->napi);
 	} else
 		e1000_irq_enable(adapter);
-#else
-	adapter->total_tx_bytes = 0;
-	adapter->total_rx_bytes = 0;
-	adapter->total_tx_packets = 0;
-	adapter->total_rx_packets = 0;
-
-	for (i = 0; i < E1000_MAX_INTR; i++)
-		if (unlikely(!adapter->clean_rx(adapter, adapter->rx_ring) &
-		   !e1000_clean_tx_irq(adapter, adapter->tx_ring)))
-			break;
-
-	if (likely(adapter->itr_setting & 3))
-		e1000_set_itr(adapter);
-#endif
 
 	return IRQ_HANDLED;
 }
@@ -3856,13 +3802,10 @@ e1000_intr(int irq, void *data)
 	struct e1000_adapter *adapter = netdev_priv(netdev);
 	struct e1000_hw *hw = &adapter->hw;
 	u32 rctl, icr = E1000_READ_REG(hw, ICR);
-#ifndef CONFIG_E1000_NAPI
-	int i;
-#endif
+
 	if (unlikely(!icr))
 		return IRQ_NONE;  /* Not our interrupt */
 
-#ifdef CONFIG_E1000_NAPI
 	/* IMS will not auto-mask if INT_ASSERTED is not set, and if it is
 	 * not set, then the adapter didn't send an interrupt */
 	if (unlikely(hw->mac_type >= e1000_82571 &&
@@ -3871,7 +3814,6 @@ e1000_intr(int irq, void *data)
 
 	/* Interrupt Auto-Mask...upon reading ICR, interrupts are masked.  No
 	 * need for the IMC write */
-#endif
 
 	if (unlikely(icr & (E1000_ICR_RXSEQ | E1000_ICR_LSC))) {
 		hw->get_link_status = 1;
@@ -3891,7 +3833,6 @@ e1000_intr(int irq, void *data)
 			mod_timer(&adapter->watchdog_timer, jiffies + 1);
 	}
 
-#ifdef CONFIG_E1000_NAPI
 	if (unlikely(hw->mac_type < e1000_82571)) {
 		/* disable interrupts, without the synchronize_irq bit */
 		E1000_WRITE_REG(hw, IMC, ~0);
@@ -3907,41 +3848,9 @@ e1000_intr(int irq, void *data)
 		/* this really should not happen! if it does it is basically a
 		 * bug, but not a hard error, so enable ints and continue */
 		e1000_irq_enable(adapter);
-#else
-	/* Writing IMC and IMS is needed for 82547.
-	 * Due to Hub Link bus being occupied, an interrupt
-	 * de-assertion message is not able to be sent.
-	 * When an interrupt assertion message is generated later,
-	 * two messages are re-ordered and sent out.
-	 * That causes APIC to think 82547 is in de-assertion
-	 * state, while 82547 is in assertion state, resulting
-	 * in dead lock. Writing IMC forces 82547 into
-	 * de-assertion state.
-	 */
-	if (hw->mac_type == e1000_82547 || hw->mac_type == e1000_82547_rev_2)
-		E1000_WRITE_REG(hw, IMC, ~0);
-
-	adapter->total_tx_bytes = 0;
-	adapter->total_rx_bytes = 0;
-	adapter->total_tx_packets = 0;
-	adapter->total_rx_packets = 0;
-
-	for (i = 0; i < E1000_MAX_INTR; i++)
-		if (unlikely(!adapter->clean_rx(adapter, adapter->rx_ring) &
-		   !e1000_clean_tx_irq(adapter, adapter->tx_ring)))
-			break;
-
-	if (likely(adapter->itr_setting & 3))
-		e1000_set_itr(adapter);
-
-	if (hw->mac_type == e1000_82547 || hw->mac_type == e1000_82547_rev_2)
-		e1000_irq_enable(adapter);
-
-#endif
 	return IRQ_HANDLED;
 }
 
-#ifdef CONFIG_E1000_NAPI
 /**
  * e1000_clean - NAPI Rx polling callback
  * @adapter: board private structure
@@ -3984,7 +3893,6 @@ e1000_clean(struct napi_struct *napi, int budget)
 	return work_done;
 }
 
-#endif
 /**
  * e1000_clean_tx_irq - Reclaim resources after transmit completes
  * @adapter: board private structure
@@ -3998,9 +3906,7 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 	struct e1000_tx_desc *tx_desc, *eop_desc;
 	struct e1000_buffer *buffer_info;
 	unsigned int i, eop;
-#ifdef CONFIG_E1000_NAPI
 	unsigned int count = 0;
-#endif
 	bool cleaned = false;
 	unsigned int total_tx_bytes=0, total_tx_packets=0;
 
@@ -4032,11 +3938,10 @@ e1000_clean_tx_irq(struct e1000_adapter *adapter,
 
 		eop = tx_ring->buffer_info[i].next_to_watch;
 		eop_desc = E1000_TX_DESC(*tx_ring, eop);
-#ifdef CONFIG_E1000_NAPI
 #define E1000_TX_WEIGHT 64
 		/* weight of a sort for tx, to avoid endless transmit cleanup */
-		if (count++ == E1000_TX_WEIGHT) break;
-#endif
+		if (count++ == E1000_TX_WEIGHT)
+			break;
 	}
 
 	tx_ring->next_to_clean = i;
@@ -4153,14 +4058,9 @@ e1000_rx_checksum(struct e1000_adapter *adapter,
  **/
 
 static bool
-#ifdef CONFIG_E1000_NAPI
 e1000_clean_rx_irq(struct e1000_adapter *adapter,
                    struct e1000_rx_ring *rx_ring,
                    int *work_done, int work_to_do)
-#else
-e1000_clean_rx_irq(struct e1000_adapter *adapter,
-                   struct e1000_rx_ring *rx_ring)
-#endif
 {
 	struct net_device *netdev = adapter->netdev;
 	struct pci_dev *pdev = adapter->pdev;
@@ -4182,11 +4082,10 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 		struct sk_buff *skb;
 		u8 status;
 
-#ifdef CONFIG_E1000_NAPI
 		if (*work_done >= work_to_do)
 			break;
 		(*work_done)++;
-#endif
+
 		status = rx_desc->status;
 		skb = buffer_info->skb;
 		buffer_info->skb = NULL;
@@ -4273,7 +4172,7 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 				  le16_to_cpu(rx_desc->csum), skb);
 
 		skb->protocol = eth_type_trans(skb, netdev);
-#ifdef CONFIG_E1000_NAPI
+
 		if (unlikely(adapter->vlgrp &&
 			    (status & E1000_RXD_STAT_VP))) {
 			vlan_hwaccel_receive_skb(skb, adapter->vlgrp,
@@ -4281,15 +4180,7 @@ e1000_clean_rx_irq(struct e1000_adapter *adapter,
 		} else {
 			netif_receive_skb(skb);
 		}
-#else /* CONFIG_E1000_NAPI */
-		if (unlikely(adapter->vlgrp &&
-			    (status & E1000_RXD_STAT_VP))) {
-			vlan_hwaccel_rx(skb, adapter->vlgrp,
-					le16_to_cpu(rx_desc->special));
-		} else {
-			netif_rx(skb);
-		}
-#endif /* CONFIG_E1000_NAPI */
+
 		netdev->last_rx = jiffies;
 
 next_desc:
@@ -4324,14 +4215,9 @@ next_desc:
  **/
 
 static bool
-#ifdef CONFIG_E1000_NAPI
 e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
                       struct e1000_rx_ring *rx_ring,
                       int *work_done, int work_to_do)
-#else
-e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
-                      struct e1000_rx_ring *rx_ring)
-#endif
 {
 	union e1000_rx_desc_packet_split *rx_desc, *next_rxd;
 	struct net_device *netdev = adapter->netdev;
@@ -4354,11 +4240,11 @@ e1000_clean_rx_irq_ps(struct e1000_adapter *adapter,
 	while (staterr & E1000_RXD_STAT_DD) {
 		ps_page = &rx_ring->ps_page[i];
 		ps_page_dma = &rx_ring->ps_page_dma[i];
-#ifdef CONFIG_E1000_NAPI
+
 		if (unlikely(*work_done >= work_to_do))
 			break;
 		(*work_done)++;
-#endif
+
 		skb = buffer_info->skb;
 
 		/* in the packet split case this is header only */
@@ -4459,21 +4345,14 @@ copydone:
 		if (likely(rx_desc->wb.upper.header_status &
 			   cpu_to_le16(E1000_RXDPS_HDRSTAT_HDRSP)))
 			adapter->rx_hdr_split++;
-#ifdef CONFIG_E1000_NAPI
+
 		if (unlikely(adapter->vlgrp && (staterr & E1000_RXD_STAT_VP))) {
 			vlan_hwaccel_receive_skb(skb, adapter->vlgrp,
 				le16_to_cpu(rx_desc->wb.middle.vlan));
 		} else {
 			netif_receive_skb(skb);
 		}
-#else /* CONFIG_E1000_NAPI */
-		if (unlikely(adapter->vlgrp && (staterr & E1000_RXD_STAT_VP))) {
-			vlan_hwaccel_rx(skb, adapter->vlgrp,
-				le16_to_cpu(rx_desc->wb.middle.vlan));
-		} else {
-			netif_rx(skb);
-		}
-#endif /* CONFIG_E1000_NAPI */
+
 		netdev->last_rx = jiffies;
 
 next_desc:
@@ -5247,9 +5126,6 @@ e1000_netpoll(struct net_device *netdev)
 
 	disable_irq(adapter->pdev->irq);
 	e1000_intr(adapter->pdev->irq, netdev);
-#ifndef CONFIG_E1000_NAPI
-	adapter->clean_rx(adapter, adapter->rx_ring);
-#endif
 	enable_irq(adapter->pdev->irq);
 }
 #endif
-- 
1.5.3.3

[PATCH 4/6] forcedeth: delete non NAPI code from the driver.

[Francois Romieu]

Compile-tested only.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 drivers/net/Kconfig     |   14 -----
 drivers/net/forcedeth.c |  145 ++---------------------------------------------
 2 files changed, 5 insertions(+), 154 deletions(-)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index 33b8899..f74461b 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -1378,20 +1378,6 @@ config FORCEDETH
 	  To compile this driver as a module, choose M here. The module
 	  will be called forcedeth.
 
-config FORCEDETH_NAPI
-	bool "Use Rx Polling (NAPI) (EXPERIMENTAL)"
-	depends on FORCEDETH && EXPERIMENTAL
-	help
-	  NAPI is a new driver API designed to reduce CPU and interrupt load
-	  when the driver is receiving lots of packets from the card. It is
-	  still somewhat experimental and thus not yet enabled by default.
-
-	  If your estimated Rx load is 10kpps or more, or if the card will be
-	  deployed on potentially unfriendly networks (e.g. in a firewall),
-	  then say Y here.
-
-	  If in doubt, say N.
-
 config CS89x0
 	tristate "CS89x0 support"
 	depends on NET_PCI && (ISA || MACH_IXDP2351 || ARCH_IXDP2X01 || ARCH_PNX010X)
diff --git a/drivers/net/forcedeth.c b/drivers/net/forcedeth.c
index 786d668..3af5e0f 100644
--- a/drivers/net/forcedeth.c
+++ b/drivers/net/forcedeth.c
@@ -1676,7 +1676,6 @@ static int nv_alloc_rx_optimized(struct net_device *dev)
 }
 
 /* If rx bufs are exhausted called after 50ms to attempt to refresh */
-#ifdef CONFIG_FORCEDETH_NAPI
 static void nv_do_rx_refill(unsigned long data)
 {
 	struct net_device *dev = (struct net_device *) data;
@@ -1685,41 +1684,6 @@ static void nv_do_rx_refill(unsigned long data)
 	/* Just reschedule NAPI rx processing */
 	netif_rx_schedule(dev, &np->napi);
 }
-#else
-static void nv_do_rx_refill(unsigned long data)
-{
-	struct net_device *dev = (struct net_device *) data;
-	struct fe_priv *np = netdev_priv(dev);
-	int retcode;
-
-	if (!using_multi_irqs(dev)) {
-		if (np->msi_flags & NV_MSI_X_ENABLED)
-			disable_irq(np->msi_x_entry[NV_MSI_X_VECTOR_ALL].vector);
-		else
-			disable_irq(np->pci_dev->irq);
-	} else {
-		disable_irq(np->msi_x_entry[NV_MSI_X_VECTOR_RX].vector);
-	}
-	if (!nv_optimized(np))
-		retcode = nv_alloc_rx(dev);
-	else
-		retcode = nv_alloc_rx_optimized(dev);
-	if (retcode) {
-		spin_lock_irq(&np->lock);
-		if (!np->in_shutdown)
-			mod_timer(&np->oom_kick, jiffies + OOM_REFILL);
-		spin_unlock_irq(&np->lock);
-	}
-	if (!using_multi_irqs(dev)) {
-		if (np->msi_flags & NV_MSI_X_ENABLED)
-			enable_irq(np->msi_x_entry[NV_MSI_X_VECTOR_ALL].vector);
-		else
-			enable_irq(np->pci_dev->irq);
-	} else {
-		enable_irq(np->msi_x_entry[NV_MSI_X_VECTOR_RX].vector);
-	}
-}
-#endif
 
 static void nv_init_rx(struct net_device *dev)
 {
@@ -2653,11 +2617,9 @@ static int nv_rx_process(struct net_device *dev, int limit)
 		skb->protocol = eth_type_trans(skb, dev);
 		dprintk(KERN_DEBUG "%s: nv_rx_process: %d bytes, proto %d accepted.\n",
 					dev->name, len, skb->protocol);
-#ifdef CONFIG_FORCEDETH_NAPI
+
 		netif_receive_skb(skb);
-#else
-		netif_rx(skb);
-#endif
+
 		dev->last_rx = jiffies;
 		dev->stats.rx_packets++;
 		dev->stats.rx_bytes += len;
@@ -2746,29 +2708,15 @@ static int nv_rx_process_optimized(struct net_device *dev, int limit)
 			dprintk(KERN_DEBUG "%s: nv_rx_process_optimized: %d bytes, proto %d accepted.\n",
 				dev->name, len, skb->protocol);
 
-			if (likely(!np->vlangrp)) {
-#ifdef CONFIG_FORCEDETH_NAPI
+			if (likely(!np->vlangrp))
 				netif_receive_skb(skb);
-#else
-				netif_rx(skb);
-#endif
-			} else {
+			else {
 				vlanflags = le32_to_cpu(np->get_rx.ex->buflow);
 				if (vlanflags & NV_RX3_VLAN_TAG_PRESENT) {
-#ifdef CONFIG_FORCEDETH_NAPI
 					vlan_hwaccel_receive_skb(skb, np->vlangrp,
 								 vlanflags & NV_RX3_VLAN_TAG_MASK);
-#else
-					vlan_hwaccel_rx(skb, np->vlangrp,
-							vlanflags & NV_RX3_VLAN_TAG_MASK);
-#endif
-				} else {
-#ifdef CONFIG_FORCEDETH_NAPI
+				} else
 					netif_receive_skb(skb);
-#else
-					netif_rx(skb);
-#endif
-				}
 			}
 
 			dev->last_rx = jiffies;
@@ -3319,7 +3267,6 @@ static irqreturn_t nv_nic_irq(int foo, void *data)
 		nv_tx_done(dev);
 		spin_unlock(&np->lock);
 
-#ifdef CONFIG_FORCEDETH_NAPI
 		if (events & NVREG_IRQ_RX_ALL) {
 			netif_rx_schedule(dev, &np->napi);
 
@@ -3333,16 +3280,6 @@ static irqreturn_t nv_nic_irq(int foo, void *data)
 				writel(np->irqmask, base + NvRegIrqMask);
 			spin_unlock(&np->lock);
 		}
-#else
-		if (nv_rx_process(dev, RX_WORK_PER_LOOP)) {
-			if (unlikely(nv_alloc_rx(dev))) {
-				spin_lock(&np->lock);
-				if (!np->in_shutdown)
-					mod_timer(&np->oom_kick, jiffies + OOM_REFILL);
-				spin_unlock(&np->lock);
-			}
-		}
-#endif
 		if (unlikely(events & NVREG_IRQ_LINK)) {
 			spin_lock(&np->lock);
 			nv_link_irq(dev);
@@ -3436,7 +3373,6 @@ static irqreturn_t nv_nic_irq_optimized(int foo, void *data)
 		nv_tx_done_optimized(dev, TX_WORK_PER_LOOP);
 		spin_unlock(&np->lock);
 
-#ifdef CONFIG_FORCEDETH_NAPI
 		if (events & NVREG_IRQ_RX_ALL) {
 			netif_rx_schedule(dev, &np->napi);
 
@@ -3450,16 +3386,6 @@ static irqreturn_t nv_nic_irq_optimized(int foo, void *data)
 				writel(np->irqmask, base + NvRegIrqMask);
 			spin_unlock(&np->lock);
 		}
-#else
-		if (nv_rx_process_optimized(dev, RX_WORK_PER_LOOP)) {
-			if (unlikely(nv_alloc_rx_optimized(dev))) {
-				spin_lock(&np->lock);
-				if (!np->in_shutdown)
-					mod_timer(&np->oom_kick, jiffies + OOM_REFILL);
-				spin_unlock(&np->lock);
-			}
-		}
-#endif
 		if (unlikely(events & NVREG_IRQ_LINK)) {
 			spin_lock(&np->lock);
 			nv_link_irq(dev);
@@ -3568,7 +3494,6 @@ static irqreturn_t nv_nic_irq_tx(int foo, void *data)
 	return IRQ_RETVAL(i);
 }
 
-#ifdef CONFIG_FORCEDETH_NAPI
 static int nv_napi_poll(struct napi_struct *napi, int budget)
 {
 	struct fe_priv *np = container_of(napi, struct fe_priv, napi);
@@ -3608,9 +3533,7 @@ static int nv_napi_poll(struct napi_struct *napi, int budget)
 	}
 	return pkts;
 }
-#endif
 
-#ifdef CONFIG_FORCEDETH_NAPI
 static irqreturn_t nv_nic_irq_rx(int foo, void *data)
 {
 	struct net_device *dev = (struct net_device *) data;
@@ -3629,54 +3552,6 @@ static irqreturn_t nv_nic_irq_rx(int foo, void *data)
 	}
 	return IRQ_HANDLED;
 }
-#else
-static irqreturn_t nv_nic_irq_rx(int foo, void *data)
-{
-	struct net_device *dev = (struct net_device *) data;
-	struct fe_priv *np = netdev_priv(dev);
-	u8 __iomem *base = get_hwbase(dev);
-	u32 events;
-	int i;
-	unsigned long flags;
-
-	dprintk(KERN_DEBUG "%s: nv_nic_irq_rx\n", dev->name);
-
-	for (i=0; ; i++) {
-		events = readl(base + NvRegMSIXIrqStatus) & NVREG_IRQ_RX_ALL;
-		writel(NVREG_IRQ_RX_ALL, base + NvRegMSIXIrqStatus);
-		dprintk(KERN_DEBUG "%s: rx irq: %08x\n", dev->name, events);
-		if (!(events & np->irqmask))
-			break;
-
-		if (nv_rx_process_optimized(dev, RX_WORK_PER_LOOP)) {
-			if (unlikely(nv_alloc_rx_optimized(dev))) {
-				spin_lock_irqsave(&np->lock, flags);
-				if (!np->in_shutdown)
-					mod_timer(&np->oom_kick, jiffies + OOM_REFILL);
-				spin_unlock_irqrestore(&np->lock, flags);
-			}
-		}
-
-		if (unlikely(i > max_interrupt_work)) {
-			spin_lock_irqsave(&np->lock, flags);
-			/* disable interrupts on the nic */
-			writel(NVREG_IRQ_RX_ALL, base + NvRegIrqMask);
-			pci_push(base);
-
-			if (!np->in_shutdown) {
-				np->nic_poll_irq |= NVREG_IRQ_RX_ALL;
-				mod_timer(&np->nic_poll, jiffies + POLL_WAIT);
-			}
-			spin_unlock_irqrestore(&np->lock, flags);
-			printk(KERN_DEBUG "%s: too many iterations (%d) in nv_nic_irq_rx.\n", dev->name, i);
-			break;
-		}
-	}
-	dprintk(KERN_DEBUG "%s: nv_nic_irq_rx completed\n", dev->name);
-
-	return IRQ_RETVAL(i);
-}
-#endif
 
 static irqreturn_t nv_nic_irq_other(int foo, void *data)
 {
@@ -4942,9 +4817,7 @@ static void nv_self_test(struct net_device *dev, struct ethtool_test *test, u64
 	if (test->flags & ETH_TEST_FL_OFFLINE) {
 		if (netif_running(dev)) {
 			netif_stop_queue(dev);
-#ifdef CONFIG_FORCEDETH_NAPI
 			napi_disable(&np->napi);
-#endif
 			netif_tx_lock_bh(dev);
 			spin_lock_irq(&np->lock);
 			nv_disable_hw_interrupts(dev, np->irqmask);
@@ -5000,9 +4873,7 @@ static void nv_self_test(struct net_device *dev, struct ethtool_test *test, u64
 			/* restart rx engine */
 			nv_start_rxtx(dev);
 			netif_start_queue(dev);
-#ifdef CONFIG_FORCEDETH_NAPI
 			napi_enable(&np->napi);
-#endif
 			nv_enable_hw_interrupts(dev, np->irqmask);
 		}
 	}
@@ -5242,9 +5113,7 @@ static int nv_open(struct net_device *dev)
 	ret = nv_update_linkspeed(dev);
 	nv_start_rxtx(dev);
 	netif_start_queue(dev);
-#ifdef CONFIG_FORCEDETH_NAPI
 	napi_enable(&np->napi);
-#endif
 
 	if (ret) {
 		netif_carrier_on(dev);
@@ -5276,9 +5145,7 @@ static int nv_close(struct net_device *dev)
 	spin_lock_irq(&np->lock);
 	np->in_shutdown = 1;
 	spin_unlock_irq(&np->lock);
-#ifdef CONFIG_FORCEDETH_NAPI
 	napi_disable(&np->napi);
-#endif
 	synchronize_irq(np->pci_dev->irq);
 
 	del_timer_sync(&np->oom_kick);
@@ -5496,9 +5363,7 @@ static int __devinit nv_probe(struct pci_dev *pci_dev, const struct pci_device_i
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	dev->poll_controller = nv_poll_controller;
 #endif
-#ifdef CONFIG_FORCEDETH_NAPI
 	netif_napi_add(dev, &np->napi, nv_napi_poll, RX_WORK_PER_LOOP);
-#endif
 	SET_ETHTOOL_OPS(dev, &ops);
 	dev->tx_timeout = nv_tx_timeout;
 	dev->watchdog_timeo = NV_WATCHDOG_TIMEO;
-- 
1.5.3.3

[PATCH 5/6] gianfar: delete non NAPI code from the driver.

[Francois Romieu]

Compile-tested only.

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 drivers/net/Kconfig   |    4 --
 drivers/net/gianfar.c |   76 ++++---------------------------------------------
 drivers/net/gianfar.h |   11 -------
 3 files changed, 6 insertions(+), 85 deletions(-)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index f74461b..c4841c0 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2234,10 +2234,6 @@ config GIANFAR
 	  This driver supports the Gigabit TSEC on the MPC83xx, MPC85xx,
 	  and MPC86xx family of chips, and the FEC on the 8540.
 
-config GFAR_NAPI
-	bool "Use Rx Polling (NAPI)"
-	depends on GIANFAR
-
 config UCC_GETH
 	tristate "Freescale QE Gigabit Ethernet"
 	depends on QUICC_ENGINE
diff --git a/drivers/net/gianfar.c b/drivers/net/gianfar.c
index 393a0f1..fa78d68 100644
--- a/drivers/net/gianfar.c
+++ b/drivers/net/gianfar.c
@@ -44,8 +44,7 @@
  *  happen immediately, but will wait until either a set number
  *  of frames or amount of time have passed).  In NAPI, the
  *  interrupt handler will signal there is work to be done, and
- *  exit.  Without NAPI, the packet(s) will be handled
- *  immediately.  Both methods will start at the last known empty
+ *  exit. This method will start at the last known empty
  *  descriptor, and process every subsequent descriptor until there
  *  are none left with data (NAPI will stop after a set number of
  *  packets to give time to other tasks, but will eventually
@@ -101,12 +100,6 @@
 #undef BRIEF_GFAR_ERRORS
 #undef VERBOSE_GFAR_ERRORS
 
-#ifdef CONFIG_GFAR_NAPI
-#define RECEIVE(x) netif_receive_skb(x)
-#else
-#define RECEIVE(x) netif_rx(x)
-#endif
-
 const char gfar_driver_name[] = "Gianfar Ethernet";
 const char gfar_driver_version[] = "1.3";
 
@@ -131,9 +124,7 @@ static void free_skb_resources(struct gfar_private *priv);
 static void gfar_set_multi(struct net_device *dev);
 static void gfar_set_hash_for_addr(struct net_device *dev, u8 *addr);
 static void gfar_configure_serdes(struct net_device *dev);
-#ifdef CONFIG_GFAR_NAPI
 static int gfar_poll(struct napi_struct *napi, int budget);
-#endif
 #ifdef CONFIG_NET_POLL_CONTROLLER
 static void gfar_netpoll(struct net_device *dev);
 #endif
@@ -260,9 +251,7 @@ static int gfar_probe(struct platform_device *pdev)
 	dev->hard_start_xmit = gfar_start_xmit;
 	dev->tx_timeout = gfar_timeout;
 	dev->watchdog_timeo = TX_TIMEOUT;
-#ifdef CONFIG_GFAR_NAPI
 	netif_napi_add(dev, &priv->napi, gfar_poll, GFAR_DEV_WEIGHT);
-#endif
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	dev->poll_controller = gfar_netpoll;
 #endif
@@ -363,11 +352,7 @@ static int gfar_probe(struct platform_device *pdev)
 
 	/* Even more device info helps when determining which kernel */
 	/* provided which set of benchmarks. */
-#ifdef CONFIG_GFAR_NAPI
 	printk(KERN_INFO "%s: Running with NAPI enabled\n", dev->name);
-#else
-	printk(KERN_INFO "%s: Running with NAPI disabled\n", dev->name);
-#endif
 	printk(KERN_INFO "%s: %d/%d RX/TX BD ring size\n",
 	       dev->name, priv->rx_ring_size, priv->tx_ring_size);
 
@@ -945,14 +930,10 @@ tx_skb_fail:
 /* Returns 0 for success. */
 static int gfar_enet_open(struct net_device *dev)
 {
-#ifdef CONFIG_GFAR_NAPI
 	struct gfar_private *priv = netdev_priv(dev);
-#endif
 	int err;
 
-#ifdef CONFIG_GFAR_NAPI
 	napi_enable(&priv->napi);
-#endif
 
 	/* Initialize a bunch of registers */
 	init_registers(dev);
@@ -962,17 +943,13 @@ static int gfar_enet_open(struct net_device *dev)
 	err = init_phy(dev);
 
 	if(err) {
-#ifdef CONFIG_GFAR_NAPI
 		napi_disable(&priv->napi);
-#endif
 		return err;
 	}
 
 	err = startup_gfar(dev);
 	if (err) {
-#ifdef CONFIG_GFAR_NAPI
 		napi_disable(&priv->napi);
-#endif
 		return err;
 	}
 
@@ -1128,9 +1105,7 @@ static int gfar_close(struct net_device *dev)
 {
 	struct gfar_private *priv = netdev_priv(dev);
 
-#ifdef CONFIG_GFAR_NAPI
 	napi_disable(&priv->napi);
-#endif
 
 	stop_gfar(dev);
 
@@ -1427,14 +1402,9 @@ irqreturn_t gfar_receive(int irq, void *dev_id)
 {
 	struct net_device *dev = (struct net_device *) dev_id;
 	struct gfar_private *priv = netdev_priv(dev);
-#ifdef CONFIG_GFAR_NAPI
 	u32 tempval;
-#else
-	unsigned long flags;
-#endif
 
 	/* support NAPI */
-#ifdef CONFIG_GFAR_NAPI
 	/* Clear IEVENT, so interrupts aren't called again
 	 * because of the packets that have already arrived */
 	gfar_write(&priv->regs->ievent, IEVENT_RTX_MASK);
@@ -1451,38 +1421,10 @@ irqreturn_t gfar_receive(int irq, void *dev_id)
 				dev->name, gfar_read(&priv->regs->ievent),
 				gfar_read(&priv->regs->imask));
 	}
-#else
-	/* Clear IEVENT, so rx interrupt isn't called again
-	 * because of this interrupt */
-	gfar_write(&priv->regs->ievent, IEVENT_RX_MASK);
-
-	spin_lock_irqsave(&priv->rxlock, flags);
-	gfar_clean_rx_ring(dev, priv->rx_ring_size);
-
-	/* If we are coalescing interrupts, update the timer */
-	/* Otherwise, clear it */
-	if (likely(priv->rxcoalescing)) {
-		gfar_write(&priv->regs->rxic, 0);
-		gfar_write(&priv->regs->rxic,
-			   mk_ic_value(priv->rxcount, priv->rxtime));
-	}
-
-	spin_unlock_irqrestore(&priv->rxlock, flags);
-#endif
 
 	return IRQ_HANDLED;
 }
 
-static inline int gfar_rx_vlan(struct sk_buff *skb,
-		struct vlan_group *vlgrp, unsigned short vlctl)
-{
-#ifdef CONFIG_GFAR_NAPI
-	return vlan_hwaccel_receive_skb(skb, vlgrp, vlctl);
-#else
-	return vlan_hwaccel_rx(skb, vlgrp, vlctl);
-#endif
-}
-
 static inline void gfar_rx_checksum(struct sk_buff *skb, struct rxfcb *fcb)
 {
 	/* If valid headers were found, and valid sums
@@ -1539,10 +1481,11 @@ static int gfar_process_frame(struct net_device *dev, struct sk_buff *skb,
 		skb->protocol = eth_type_trans(skb, dev);
 
 		/* Send the packet up the stack */
-		if (unlikely(priv->vlgrp && (fcb->flags & RXFCB_VLN)))
-			ret = gfar_rx_vlan(skb, priv->vlgrp, fcb->vlctl);
-		else
-			ret = RECEIVE(skb);
+		if (unlikely(priv->vlgrp && (fcb->flags & RXFCB_VLN))) {
+			ret = vlan_hwaccel_receive_skb(skb, priv->vlgrp,
+						       fcb->vlctl);
+		} else
+			ret = netif_receive_skb(skb);
 
 		if (NET_RX_DROP == ret)
 			priv->extra_stats.kernel_dropped++;
@@ -1629,7 +1572,6 @@ int gfar_clean_rx_ring(struct net_device *dev, int rx_work_limit)
 	return howmany;
 }
 
-#ifdef CONFIG_GFAR_NAPI
 static int gfar_poll(struct napi_struct *napi, int budget)
 {
 	struct gfar_private *priv = container_of(napi, struct gfar_private, napi);
@@ -1664,7 +1606,6 @@ static int gfar_poll(struct napi_struct *napi, int budget)
 
 	return howmany;
 }
-#endif
 
 #ifdef CONFIG_NET_POLL_CONTROLLER
 /*
@@ -2003,11 +1944,6 @@ static irqreturn_t gfar_error(int irq, void *dev_id)
 
 		gfar_receive(irq, dev_id);
 
-#ifndef CONFIG_GFAR_NAPI
-		/* Clear the halt bit in RSTAT */
-		gfar_write(&priv->regs->rstat, RSTAT_CLEAR_RHALT);
-#endif
-
 		if (netif_msg_rx_err(priv))
 			printk(KERN_DEBUG "%s: busy error (rstat: %x)\n",
 			       dev->name, gfar_read(&priv->regs->rstat));
diff --git a/drivers/net/gianfar.h b/drivers/net/gianfar.h
index 27f37c8..bead71c 100644
--- a/drivers/net/gianfar.h
+++ b/drivers/net/gianfar.h
@@ -77,13 +77,8 @@ extern const char gfar_driver_name[];
 extern const char gfar_driver_version[];
 
 /* These need to be powers of 2 for this driver */
-#ifdef CONFIG_GFAR_NAPI
 #define DEFAULT_TX_RING_SIZE	256
 #define DEFAULT_RX_RING_SIZE	256
-#else
-#define DEFAULT_TX_RING_SIZE    64
-#define DEFAULT_RX_RING_SIZE    64
-#endif
 
 #define GFAR_RX_MAX_RING_SIZE   256
 #define GFAR_TX_MAX_RING_SIZE   256
@@ -128,14 +123,8 @@ extern const char gfar_driver_version[];
 
 #define DEFAULT_RXTIME	21
 
-/* Non NAPI Case */
-#ifndef CONFIG_GFAR_NAPI
-#define DEFAULT_RX_COALESCE 1
-#define DEFAULT_RXCOUNT	16
-#else
 #define DEFAULT_RX_COALESCE 0
 #define DEFAULT_RXCOUNT	0
-#endif /* CONFIG_GFAR_NAPI */
 
 #define MIIMCFG_INIT_VALUE	0x00000007
 #define MIIMCFG_RESET           0x80000000
-- 
1.5.3.3

[PATCH 6/6] ucc_geth: delete non NAPI code from the driver.

[Francois Romieu]

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
---
 drivers/net/Kconfig    |    4 ----
 drivers/net/ucc_geth.c |   29 ++---------------------------
 2 files changed, 2 insertions(+), 31 deletions(-)

diff --git a/drivers/net/Kconfig b/drivers/net/Kconfig
index c4841c0..cc78e79 100644
--- a/drivers/net/Kconfig
+++ b/drivers/net/Kconfig
@@ -2242,10 +2242,6 @@ config UCC_GETH
 	  This driver supports the Gigabit Ethernet mode of the QUICC Engine,
 	  which is available on some Freescale SOCs.
 
-config UGETH_NAPI
-	bool "Use Rx Polling (NAPI)"
-	depends on UCC_GETH
-
 config UGETH_MAGIC_PACKET
 	bool "Magic Packet detection support"
 	depends on UCC_GETH
diff --git a/drivers/net/ucc_geth.c b/drivers/net/ucc_geth.c
index fb0b918..07933f7 100644
--- a/drivers/net/ucc_geth.c
+++ b/drivers/net/ucc_geth.c
@@ -3500,11 +3500,7 @@ static int ucc_geth_rx(struct ucc_geth_private *ugeth, u8 rxQ, int rx_work_limit
 
 			dev->stats.rx_bytes += length;
 			/* Send the packet up the stack */
-#ifdef CONFIG_UGETH_NAPI
 			netif_receive_skb(skb);
-#else
-			netif_rx(skb);
-#endif				/* CONFIG_UGETH_NAPI */
 		}
 
 		ugeth->dev->last_rx = jiffies;
@@ -3580,7 +3576,6 @@ static int ucc_geth_tx(struct net_device *dev, u8 txQ)
 	return 0;
 }
 
-#ifdef CONFIG_UGETH_NAPI
 static int ucc_geth_poll(struct napi_struct *napi, int budget)
 {
 	struct ucc_geth_private *ugeth = container_of(napi, struct ucc_geth_private, napi);
@@ -3607,7 +3602,6 @@ static int ucc_geth_poll(struct napi_struct *napi, int budget)
 
 	return howmany;
 }
-#endif				/* CONFIG_UGETH_NAPI */
 
 static irqreturn_t ucc_geth_irq_handler(int irq, void *info)
 {
@@ -3617,9 +3611,6 @@ static irqreturn_t ucc_geth_irq_handler(int irq, void *info)
 	struct ucc_geth_info *ug_info;
 	register u32 ucce;
 	register u32 uccm;
-#ifndef CONFIG_UGETH_NAPI
-	register u32 rx_mask;
-#endif
 	register u32 tx_mask;
 	u8 i;
 
@@ -3636,21 +3627,11 @@ static irqreturn_t ucc_geth_irq_handler(int irq, void *info)
 
 	/* check for receive events that require processing */
 	if (ucce & UCCE_RX_EVENTS) {
-#ifdef CONFIG_UGETH_NAPI
 		if (netif_rx_schedule_prep(dev, &ugeth->napi)) {
 			uccm &= ~UCCE_RX_EVENTS;
 			out_be32(uccf->p_uccm, uccm);
 			__netif_rx_schedule(dev, &ugeth->napi);
 		}
-#else
-		rx_mask = UCCE_RXBF_SINGLE_MASK;
-		for (i = 0; i < ug_info->numQueuesRx; i++) {
-			if (ucce & rx_mask)
-				ucc_geth_rx(ugeth, i, (int)ugeth->ug_info->bdRingLenRx[i]);
-			ucce &= ~rx_mask;
-			rx_mask <<= 1;
-		}
-#endif /* CONFIG_UGETH_NAPI */
 	}
 
 	/* Tx event processing */
@@ -3720,9 +3701,8 @@ static int ucc_geth_open(struct net_device *dev)
 		return err;
 	}
 
-#ifdef CONFIG_UGETH_NAPI
 	napi_enable(&ugeth->napi);
-#endif
+
 	err = ucc_geth_startup(ugeth);
 	if (err) {
 		if (netif_msg_ifup(ugeth))
@@ -3783,9 +3763,8 @@ static int ucc_geth_open(struct net_device *dev)
 	return err;
 
 out_err:
-#ifdef CONFIG_UGETH_NAPI
 	napi_disable(&ugeth->napi);
-#endif
+
 	return err;
 }
 
@@ -3796,9 +3775,7 @@ static int ucc_geth_close(struct net_device *dev)
 
 	ugeth_vdbg("%s: IN", __FUNCTION__);
 
-#ifdef CONFIG_UGETH_NAPI
 	napi_disable(&ugeth->napi);
-#endif
 
 	ucc_geth_stop(ugeth);
 
@@ -4050,9 +4027,7 @@ static int ucc_geth_probe(struct of_device* ofdev, const struct of_device_id *ma
 	dev->hard_start_xmit = ucc_geth_start_xmit;
 	dev->tx_timeout = ucc_geth_timeout;
 	dev->watchdog_timeo = TX_TIMEOUT;
-#ifdef CONFIG_UGETH_NAPI
 	netif_napi_add(dev, &ugeth->napi, ucc_geth_poll, UCC_GETH_DEV_WEIGHT);
-#endif				/* CONFIG_UGETH_NAPI */
 #ifdef CONFIG_NET_POLL_CONTROLLER
 	dev->poll_controller = ucc_netpoll;
 #endif
-- 
1.5.3.3

RE: [PATCH 3/6] e1000: delete non NAPI code from the driver.

[Kirsher, Jeffrey T]

>-----Original Message-----
>From: Francois Romieu [mailto:romieu@fr.zoreil.com]
>Sent: Thursday, July 10, 2008 3:32 PM
>To: David Miller
>Cc: jeff@garzik.org; netdev@vger.kernel.org; Kirsher, Jeffrey T;
>Brandeburg, Jesse; Allan, Bruce W; Waskiewicz Jr, Peter P; Ronciak,
John
>Subject: [PATCH 3/6] e1000: delete non NAPI code from the driver.
>
>Compile-tested only.
>
>Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
>---
> Documentation/networking/e1000.txt |   14 +---
> drivers/net/Kconfig                |   14 ----
> drivers/net/e1000/e1000.h          |   11 +--
> drivers/net/e1000/e1000_main.c     |  150
+++-----------------------------
>---
> 4 files changed, 17 insertions(+), 172 deletions(-)
>

I will pull this into my e1000 queue.  I have several other e1000
cleanup patches also, so I put push this out with my other patches
before this weekend.

Cheers,
Jeff

RE: [PATCH 6/6] ucc_geth: delete non NAPI code from the driver.

[Li Yang]

> -----Original Message-----
> From: Francois Romieu [mailto:romieu@fr.zoreil.com] 
> Sent: Friday, July 11, 2008 6:35 AM
> To: David Miller
> Cc: jeff@garzik.org; netdev@vger.kernel.org; Fleming Andy; Li Yang
> Subject: [PATCH 6/6] ucc_geth: delete non NAPI code from the driver.
> 
> Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>

I thought NAPI is left as an option because it adds the latency.  Is it
not the case any more?

- Leo

Re: [PATCH 1/6] cxgb: delete non NAPI code from the driver.

[Jeff Garzik]

Francois Romieu wrote:
> Compile-tested only.
> 
> Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
> ---
>  drivers/net/Kconfig         |    8 -----
>  drivers/net/chelsio/cxgb2.c |    2 -
>  drivers/net/chelsio/sge.c   |   70 +++---------------------------------------
>  3 files changed, 5 insertions(+), 75 deletions(-)

applied:  cxgb (this patch), via-rhine, gianfar, ucc_geth